HV Power Supply failues

[JBro63]

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

[Max DN]

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:

--
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 neonixie-l+...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/a93fbd9d-36aa-4791-bffa-54f7d2514cdfn%40googlegroups.com.

[JBro63]

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.

[Max DN]

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).

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/513d8a2a-cb3e-473e-a91f-45e6d50bb56cn%40googlegroups.com.

[Richard Scales]

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

[Jeff Walton]

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 

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/2fe0e94c-244f-494b-938c-db201d6fdf50n%40googlegroups.com.

[gregebert]

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.

[JBro63]

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

[newxito]

Having to add a 470uF electrolytic somewhat defeats the purpose of such a small power supply. I only used this PSU once for a flat clock base and it's working perfectly so far. I used 4 x 47uF MLCCs in parallel to keep the hole thing flat.

[Richard Scales]

I would tend to agree with the thought that using 12V means it has to work less though it seems the various spikes that can come from a 12V wall wart can sometimes be all too much - hence the cap (as well as optional TVS etc as Jeff suggested).

I use the Microchip HV5XXX drivers like HV5522 etc which demand a 12V supply and logic levels (Absolutely NOT my intention to start that particular discussion regarding using them at lower logic levels!).

I cant really comment on the protection of the output - my suggestions are based purely on the recommendation of the original designer. It is a fact that I have never experienced any issue when using them with a 5V supply and I guess I could pick up the 5V rail from the 5V buck converter that I already have installed but to go from 12V to 5V just to get back to 170V seems somewhat counter intuitive.

- RIchard

[newxito]

By the way, my attempt to make a similarly flat nixie power supply failed miserably. I tried using the CJ5143-ALC flyback transformer which is 4.6mm high. It worked somehow but I was never happy with my design.

[David Pye]

I'm moving to try to power mine from usb-c as there are a range of autonegotiation boards that can request supply voltages from 5-24volt from a suitable PD PSU.

David

On Mon, 16 Feb 2026, 14:08 newxito, <axt...@gmail.com> wrote:

By the way, my attempt to make a similarly flat nixie power supply failed miserably. I tried using the CJ5143-ALC flyback transformer which is 4.6mm high. It worked somehow but I was never happy with my design.

--

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 neonixie-l+...@googlegroups.com.

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/2976622b-1c50-4cf0-804a-6a3d37c95346n%40googlegroups.com.

[Richard Scales]

I've just tried one of those and for sure it works though I don't think that USB-C with PD capability is as ubiquitous as some suggest. 

For myself I prefer to distribute 12v from  beefy 10A supplies to individual clocks but that's only really because of the number I like to display. 

For regular folk I can see that  USB-C-PD could work well. 

- Richard

You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/myKhY_ZZl4Y/unsubscribe.
To unsubscribe from this group and all its topics, send an email to neonixie-l+...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/CAOQ6x0F8Gm6prhsG6CYEdEh8NOi-dS94quk6-D%2BXh%2BH_jCsXkg%40mail.gmail.com.

[Tomasz Kowalczyk]

What is the brand of the power supply you are using for this clock? 

Was the same wall wart used during both failures? 

Is it the same model of wall wart as your other clocks are using?

Do you run another clock which has exactly the same PCB and design as the device of concern?

Do you own an oscilloscope to look at the 12 V rail?

I suspect the issue is just failing/poor quality wall wart. I've seen cheap 12 V LED power supplies create spikes exceeding 16 V. Some time later the 16 V rated capacitor inside the PSU exploded. In your case, the NCH8200HV gave in first, for one reason and another. The absurdally large capacitor the designer recommended is meant to filter any garbage poor quality wall warts generate, but I think it's smarter to just invest in reputable wall warts and skip all of this nonsense altogether.  My recommendation are Mean Well power supplies, they are not only reliable, but also electrically quiet. I used mostly GSM series. 

[David Pye]

Agree - one convenient source of PD USB C power supplies that are almost guaranteed to be suitable are the laptop USB-C PSUs.

eg Dell or Lenovo thinkpad style.  

All the ones I've seen are PD and capable of 20V (but will also supply lower voltages).

David

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/CA%2B9r3s_BfguRyPpGKgtkXze19gOx3NokYyVgwwTyOo7o13%2BhqQ%40mail.gmail.com.

[JBro63]

What is the brand of the power supply you are using for this clock? 

Was the same wall wart used during both failures? 

This was from an old BT router I'd hoarded I think, 12v 1.2A, branded I.T.E. It was the same PSU in both failures. Output is 13.8v so lesson learned not to make assumptions about these things and to check in future before use.

Is it the same model of wall wart as your other clocks are using?

No, others use mix of 12v 1A for the IN-12s (4 & 6 tube versions) and 12v 2A for an IN-18 (6 tube version) - bought new recently unbranded from Amazon. I've measured these and they're ok for 12v. Not checked for spikes yet. The IN-18 uses the bigger Omnixie 6300.

Do you run another clock which has exactly the same PCB and design as the device of concern?

Yes, I've had a build frenzy recently and have 5 identical IN-12 / K155ID1 / 8200 models. It's 2 of these that I've had the failure on - both were powered by the exBT PSU.

Do you own an oscilloscope to look at the 12 V rail?

Yes, a very old one. I have recently bought new probes so will be a good excuse to fire it up / learn how to use it. Should this be measured under load?

I've added the capacitors as suggested above. Still awaiting the TVS diodes. Luckily there is enough room in the cases for them to be added unseen. One other benefit of the capacitors has been reduction in high pitched whine the HV boards emit.

Thanks for the tip the PSUs. Farnell stock them

Thanks again all.

[Paul Andrews]

I use my own designed converters and it’s real nice knowing how the supply works so you can diagnose issues like this. I’ve used the design in a lot of clocks and never had a problem except in one clock which, after a while, started showing signs of distress - tubes not being full brightness, flickering, noise etc. my converter can be driven with anything from 3V to 24V, but this one was using an apple 5V USB adapter rated at 2.1A. Turned out the adapter was flaking out on me, which was a relief. I didn’t want to discover an issue with my design! Anyway, the moral is: Don’t trust the adapters, but also start the diagnosis with the stuff that is easy to change before you open up the case and get in to the electronics.

[Paul Andrews]

Re. PD will warts. Your clock would need to be able to negotiate the correct voltage. 5V just takes a couple of resistors. AFAIK anything else you would need a chip for.

[Jeff Walton]

Agree on not blindly trusting the 12v adaptors.

 

If you have clocks that use the NCH6300HV or NCH8200HV supplies, the issue is that they WILL die catastrophically if they see inputs over 15v.  And they will potentially take out other elements in the neighborhood when they go.  Check the 6300 datasheets and 8200 datasheets for the limits. 

There are a lot of clocks that use them.  Often, the clocks that use them don’t have much (or any) filtering or input protection on the board and just assume that the 12v power supply that is used will have sufficient filtering/protection.  That’s the flaw.  Not all wall warts or bricks are created equal and they may seem to be working fine, right up until they aren’t.  You may get lucky and notice flickering or one day you may see nothing working.  The additional components that were suggested are insurance for when the PS fails.  And they do fail.  Even though a crowbar circuit serves no functional purpose for the operation of the power supply, it is there to protect everything else.  This is just a simple response when the clock has no protection on the input and is easy to add.  They can be added right at the input jack.  Choose a different cap size if this is “too big”.

 

The 6300HV and 8200HV are great little converters, small, efficient and they power lots of cool nixie and panaplex clocks.  Don’t skimp on input protection and don’t count on the fact that the 12v PS is always going to be perfect.  If the clock can be designed to run on 5v and uses the 6300HV or 8200HV converters, you are better off using 5v because there is more margin for safety than using 12v.  If you need to have level conversions and need 12v, then protect yourself with some filtering and a TVS.

 

Just my thoughts…

 

Jeff

 

 

 

From: neoni...@googlegroups.com <neoni...@googlegroups.com> On Behalf Of Paul Andrews
Sent: Saturday, February 21, 2026 10:03 AM
To: neonixie-l <neoni...@googlegroups.com>
Subject: Re: [neonixie-l] HV Power Supply failues

 

I use my own designed converters and it’s real nice knowing how the supply works so you can diagnose issues like this. I’ve used the design in a lot of clocks and never had a problem except in one clock which, after a while, started showing signs of distress - tubes not being full brightness, flickering, noise etc. my converter can be driven with anything from 3V to 24V, but this one was using an apple 5V USB adapter rated at 2.1A. Turned out the adapter was flaking out on me, which was a relief. I didn’t want to discover an issue with my design! Anyway, the moral is: Don’t trust the adapters, but also start the diagnosis with the stuff that is easy to change before you open up the case and get in to the electronics.

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/9c3c0632-c473-4b38-8dee-fa84d8166e71n%40googlegroups.com.

[newxito]

This is the input protection I'm using for my nixie devices. As always use at your own risk.

Probably overkill...