Re: [neonixie-l] Erratic Harrison clock

[Mich...@aol.com]

I have about 10 of the clocks.

 

Various issues with them.

Mostly the issue was with a 4017 (certain type)

It would count to the 10th hour and reset instead of going to 11, 12

 

Sounds like you might have some line noise.  I believe the original circuit was changed to add some caps.  Do you have the circuit you used?

 

 

Michail Wilson
206-920-6312

 

In a message dated 3/16/2014 1:39:24 P.M. Pacific Daylight Time, brian...@gmail.com writes:

Hi all,

 What could make one of the Mike Harrison Nixie clocks run erratically? It's basically running fast, but it also does this erratically- sometimes it looks like the seconds are counting at the proper speed, but then they will speed up...?

 I'm using the pre-made board with an added seconds driver board, and it's a new build; just testing it at this time.

 Thanks for any ideas,

-Brian

[threeneurons]

Good point. Standard practice with logic ICs is to distribute 0.01 to 0.1uf ceramic caps uniformly around the board, from  power to ground. I like to use one cap per IC. That's why I recommend people should get an old copy of the TTL Cookbook (from Don Lancaster). CMOS Cookbook is also good. 4000 CMOS, as used on the Harrison clock should be more immune to noise than 7400 TTL. So If you use the TTL precautions on CMOS, it should give you extra insurance. 

Also, make sure the AC, used as the timebase is clean. This circuit uses high value resistor, and lets the 4013's static protection, clamp the voltage. There's also a small capacitor to filter the line noise.  

[petehand]

On Sunday, March 16, 2014 10:41:54 PM UTC-7, threeneurons wrote:

This circuit uses high value resistor, and lets the 4013's static protection, clamp the voltage.

Surely you cannot be serious!

Haha, threeneurons, you had me there. For a moment I actually believed someone might be ignorant enough to do it that way.

[Nick]

Being fair, this clock was designed a long long time ago when stuff directly connected to the mains was not so frowned upon - Mike is a professional engineer and his page is covered with notes recommending that isolation transformers are used - he readily admits that today, he would never design a clock like this.

There is pretty much no immunity from mains-borne noise in the circuit - if you have anything at all noisy on the same ring/whatever, your clock will run fast or the noise will cause all sorts of other exciting issues.

The circuit absolutely must have 100n ceramics for each chip (standard practice anyway) instead of just the couple it uses, and the 5V supply is derived direct from the mains via a 4V7 zener/100uF, i.e. no noise filtering there either. The 50/60Hz mains input is current limited by 2Mohm and then the ESD protection of a 4013 used as a buffer is used to clean the signal a bit. If you want mains-derived 50/60Hz then you should really use a band-pass filter as nowadays there is all sorts of signalling superimposed on the mains by the utility providers and others, e.g. 1KHz tones for switching to night-time ("Economy 7" in the UK) etc...

Nuff said, really.

Personally, I'd never recommend someone build this circuit - there are far better & safer ones out there.

Nick

[Adam Jacobs]

I can tell you that if you want to advance one of these clocks in your house hours at a time by remote control, all you need to do is work some CW in the basement. :D

-Adam

On 3/17/2014 2:38 AM, Nick wrote:

Being fair, this clock was designed a long long time ago when stuff directly connected to the mains was not so frowned upon - Mike is a professional engineer and his page is covered with notes recommending that isolation transformers are used - he readily admits that today, he would never design a clock like this.

There is pretty much no immunity from mains-borne noise in the circuit - if you have anything at all noisy on the same ring/whatever, your clock will run fast or the noise will cause all sorts of other exciting issues.

The circuit absolutely must have 100n ceramics for each chip (standard practice anyway) instead of just the couple it uses, and the 5V supply is derived direct from the mains via a 4V7 zener/100uF, i.e. no noise filtering there either. The 50/60Hz mains input is current limited by 2Mohm and then the ESD protection of a 4013 used as a buffer is used to clean the signal a bit.

Nuff said, really.

Personally, I'd never recommend someone build this circuit - there are far better & safer ones out there.

Nick

--
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 post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/c9c1566f-0013-47e5-ba6b-27fe1fa96e89%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[threeneurons]

Actually, there's a Microchip app note, that describes the very same technique. Personally, I don't like using it. I have a lot of respect for Mike Harrison. He's a real sharp guy, and I'm really thankful for his website ! As Nick mentioned, this is a really old design, and Mike just probably whipped it up, quickly. He probably never imagined that nixie tubes would become as big as a fad, as they are. He might have designed it differently, had he known that more people, than just other geeky engineering types, would be trying to build it. I got into the nixies, in the summer of 2003, and the hobby (and this clock design) had already been in existence a few years. Who knew, it would have this staying power !?

[Mike Harrison]

On Mon, 17 Mar 2014 07:44:03 -0700 (PDT), you wrote:

>Actually, there's a Microchip app note, that describes the very same
>technique. Personally, I don't like using it. I have a lot of respect for
>Mike Harrison. He's a real sharp guy, and I'm really thankful for his
>website ! As Nick mentioned, this is a really old design, and Mike just
>probably whipped it up, quickly. He probably never imagined that nixie
>tubes would become as big as a fad, as they are. He might have designed it
>differently, had he known that more people, than just other geeky
>engineering types, would be trying to build it. I got into the nixies, in
>the summer of 2003, and the hobby (and this clock design) had already been
>in existence a few years. Who knew, it would have this staying power !?
>
>On Monday, March 17, 2014 1:11:15 AM UTC-7, petehand wrote:
>>
>> On Sunday, March 16, 2014 10:41:54 PM UTC-7, threeneurons wrote:
>>>

>>> This circuit uses high value resistor, and *lets the 4013's static
>>> protection, clamp the voltage*.

>>>
>>
>> Surely you cannot be serious!
>>
>> Haha, threeneurons, you had me there. For a moment I actually believed
>> someone might be ignorant enough to do it that way.
>>

Yes, it can cause minor issues on microcontrollers with analogue circuitry due to current flowing
in paths that can cause unwanted offsets, but for simple logic, provided that the source current is
sensibly limited, it generally works just fine, though probably a good idea to use it on a schmitt
trigger input to guarantee a clean switch. I only recall a couple of reports of erratic timing over
the years.
A large (few meg) resistor between S and Q might help by adding some hysteresis.

Yes, it was designed pretty quickly, and there are other aspects of that 15-year-old design I
certainly would do differently, like the dodgy, timing-sensitive hours reset logic, having the
seconds dividers in the wrong order and the slow time-setting, but the mains supply and 50Hz input
would be low on the list of changes. I like the lack of a lumpy mains supply, which allows for some
very compact constructions.

I might have published an alternate LVAC or DC+xtal option as alternatives if I'd realised how
popular it would become.

The original goal was something small, simple and not using a microcontroller, which at the time
was relatively expensive, and needed a programmer, which many people wouldn't have had.

My biggest regret is that I didn't buy a big stock of tubes *which were going for peanuts at the
time) to do some kits, or sell PCBs.

[Nick]

On Monday, 17 March 2014 14:35:21 UTC, Adam Jacobs wrote:

I can tell you that if you want to advance one of these clocks in your house hours at a time by remote control, all you need to do is work some CW in the basement. :D

Now that the mains (in the UK at least) has no guaranteed long-term accuracy (UK does not subscribe to UCTE), it makes far more sense to use a 4060 & a 32,768Hz watch xtal - that gives you a 2Hz output that you can use a latch to divide by two to 1Hz - much more reliable... or to use a DS3231 for even greater accuracy...

I spoke to mike about this clock some years back at a UK Teslathon - that's when he commented about not doing a design like this nowadays - he's a serious professional engineer (have you seen his massive polychromatic UFO flying over Rio de Janeiro & Gdansk? http://www.dezeen.com/2008/07/15/ufo-by-peter-coffin-and-cinimod-studio/)

Nick

[threeneurons]

Yeah, toobs. I really screwed the pooch on that one, too. Didn't buy enough, when the prices where low. :o(

~1990, I started using the HC05 microcontroller. The evaluation board, which could be used as an emulator/debugger/programmer cost $500, and I thought that was a bargain price ! I had used earlier systems that cost in excess of $10,000 ! Even by 2000, a reprogrammable HC05-C8 (UV window) cost ~$50. Now, you can get in the game with ISPs for ~$35, and parts under a buck !

Still a great site:

http://www.electricstuff.co.uk/

On Monday, March 17, 2014 8:04:32 AM UTC-7, mikeselectricstuff wrote:

On Mon, 17 Mar 2014 07:44:03 -0700 (PDT), you wrote:


Yes, it was designed pretty quickly, and there are other aspects of that 15-year-old design I

certainly would do differently, ...

[chuck richards]

Let us not forget the Dallas/Maxim DS32KHZN/DIP TXCO.
That puts out a fairly accurate 32.768 kHz, guaranteed
within +/- 2 ppm.

For much greater accuracy, I've devised a programmable
"pulse-remover" which shaves off a predetermined amount
of pulses each hour. This, because every one of those
TCXOs I've encountered tends to run a bit fast.

I've got the remover removing 110 pulses per hour
in one clock right now, and it's been free-running since
Oct. 17, 2013, and presently at this time, it's lost
about half a second. Not bad for 5 months free-running.

Chuck

>>Nick
>>
>>--
>>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 post to this group, send an email to neoni...@googlegroups.com.

>>To view this discussion on the web, visit

>https://groups.google.com/d/msgid/neonixie-l/6f7dda5b-f121-4e41-a673-
>d352acfefa3d%40googlegroups.com.

>>For more options, visit https://groups.google.com/d/optout.
>>

$4.95/mo. National Dialup, Anti-Spam, Anti-Virus, 5mb personal web space. 5x faster dialup for only $9.95/mo. No contracts, No fees, No Kidding! See http://www.All2Easy.net for more details!

[Nick]

On Monday, 17 March 2014 15:46:33 UTC, Chuck wrote:

...

For much greater accuracy, I've devised a programmable
"pulse-remover" which shaves off a predetermined amount
of pulses each hour.  This, because every one of those
TCXOs I've encountered tends to run a bit fast.

 

Essentially this is a digital low-pass filter - Ronald Dekker does the same on his nixie clocks some of which use almost the same way of getting the mains frequency - you can also determine the fundamental, i.e. is the clock running on 50 or 60Hz mains, and adjust accordingly...

Nick

[chuck richards]

The way the pulse-remover works is at the end of
each hour it gets triggered, which shuts off the
32.768 kHz pulses from getting to the first
divider stage of the clock.

A binary counter then counts out the "removed" pulses
which are entered on an 8-bit dip switch.

At the end of the counting, a flip-flop is clocked which
returns the pulse stream to normal.

So far in my experiments with this "remover" which I have
on 4 clocks here, it averages out to around 110 to 115
pulses which need to be removed each hour. Each individual
TCXO is a bit different.

If I ever encounter a slow TCXO, I've got some ideas
for making a "pulse-stuffer" instead.

Chuck
>
>
>---- Original Message ----
>From: ni...@desmith.net
>To: neoni...@googlegroups.com
>Subject: Re: [neonixie-l] Erratic Harrison clock

>>--
>>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 post to this group, send an email to neoni...@googlegroups.com.
>>To view this discussion on the web, visit

>https://groups.google.com/d/msgid/neonixie-l/4e403071-a508-4b88-841c-
>ac7bfca8bdea%40googlegroups.com.

[Blitzen]

Thanks guys, I guess...I was hoping there might be a solution or something in particular to check. 

What I don't get is that this is the third of these that I've made, and the other two had no problems at all. In fact, I took this one and plugged it in next to one that's been running (with perfect accuracy) for years, and this new build displayed the same fast-running and erratic behavior. I bought parts for all these at the same time, too- no difference there although maybe there are manufacturing variations?

Thanks,

-Brian

[David Forbes]

Brian,

If you socketed the ICs, you could try swapping them between the clocks.

If you have access to an oscilloscope, you can look at the signals into and out
of the mains digitizer circuit (the one getting all the scrutiny in this
discussion) so look for differences.

--David

[chuck richards]

Sounds to me like a classic problem when using
digital logic to do stuff for which it was not intended
to do. Operating them outside of published specifications
always leads to erratic performance.

Just my two cents worth.

Chuck
>
>
>---- Original Message ----
>From: brian...@gmail.com
>To: neoni...@googlegroups.com
>Subject: Re: [neonixie-l] Erratic Harrison clock

>Date: Mon, 17 Mar 2014 15:34:24 -0700 (PDT)
>
>>Thanks guys, I guess...I was hoping there might be a solution or
>something
>>in particular to check.
>>What I don't get is that this is the third of these that I've made,
>and the
>>other two had no problems at all. In fact, I took this one and
>plugged it
>>in next to one that's been running (with perfect accuracy) for
>years, and
>>this new build displayed the same fast-running and erratic behavior.
>I
>>bought parts for all these at the same time, too- no difference
>there
>>although maybe there are manufacturing variations?
>>Thanks,
>>-Brian
>>

>>--
>>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 post to this group, send an email to neoni...@googlegroups.com.
>>To view this discussion on the web, visit

>https://groups.google.com/d/msgid/neonixie-l/9e8e267f-826d-4e30-827b-
>15b5ca44fb91%40googlegroups.com.

[Nick]

In many cases like this I've found that different manufacturers' equivalents/compatibles behave differently when under pressure or at an edge condition.. e.g. CD4013BC, HEF4013B, HCF4013, MC14013B etc. 

Are you using the same manufacturer's devices in this clock as the other ones that do work correctly?

Nick

[Ian Vine]

Hi all,

Many years ago I worked as a product engineer in a semi custom silicon fab (Plessey in Oldham, uk)

The job was to troubleshoot problems with the final product. A lot of the time the fault was with the end customer using the device out of spec. Prototype 1, 2 and 3 work. Production run of 1000 gave a high failure rate.

Further to this you will get variations from batch to batch from the same manufacturer. Variations in substrate, the doping process and so on. 

IanV

--

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 post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/926e0faa-609c-4765-8107-f4290e1989ad%40googlegroups.com.

[threeneurons]

On Tuesday, March 18, 2014 3:22:47 AM UTC-7, iavine wrote:

 A lot of the time the fault was with the end customer using the device out of spec. Prototype 1, 2 and 3 work. Production run of 1000 gave a high failure rate.

 One of the common sayings, of one of my old bosses was, "there is nothing more treacherous than a working prototype".

I remember one that he made, and tested, then had a limited run of 25 made. They failed. The prototype had a bad diode, that was shorted. So, in this case the fix was simple.

I'm not too keen on that Microchip App Note, either. But they published it. Personally, I use a buffer stage (NPN xstr) with a simple RC filter. If it goes into a uC input, then I do further filtering in software. Usually, a variation of switch debounce routine. I always like using buffer stages. If nothing more than just something "cheap" to blowout first, before something more expensive, with more leads to desolder, pops.

[Terry S]

What in the original post led you to conclude that he is "operating them outside the published specs"?

I've operated TTL and CMOS out of spec many times.... it doesn't ALWAYS lead to erratic performance. You have to pick and choose which specs to bend. For instance, standard TTL fanout is10 UL. But everyone knows you can typically get away with twice that. Depends on your circuit.

Terry

[Mich...@aol.com]

CD4013BC, HEF4013B, HCF4013, MC14013B etc

 

That was my issue.

 

When I get time, I will pull some of those clocks up and see which were working correctly and which do not.  I do recall searching ebay for the specific 4013 that did work and purchased; however, the seller was using the picture of the one that did work, but didn't send them.

 

I will check this weekend if you would like, but if you have both the exact working and and exact not working in front of you, it would be easier for you to compare.

 

Also to note, the circuit boards I purchased were exactly the same; however, the caps on the mains was diff as well.

 

You have pictures of the clocks online?

 

Michail Wilson
206-920-6312

 

In a message dated 3/18/2014 2:45:11 A.M. Pacific Daylight Time, ni...@desmith.net writes:

In many cases like this I've found that different manufacturers' equivalents/compatibles behave differently when under pressure or at an edge condition.. e.g. CD4013BC, HEF4013B, HCF4013, MC14013B etc. 

Are you using the same manufacturer's devices in this clock as the other ones that do work correctly?

Nick

--

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 post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/926e0faa-609c-4765-8107-f4290e1989ad%40googlegroups.com.