TheRadioBoard

I'm putting together something I have wanted to build since I was a kid - a nixie clock. I purchased 4 NOS IN16 nixies and Russian equivalents of the 74141 driver IC and I'm just in the process of building up a test power supply for the HT side (Nixies require around 170V at around 2mA to light them up). The circuit I am using is here: http://www.ledsales.com.au/kits/nixie_supply.pdf

I purchased all the bits I didn't have, without really thinking much more than, "I have 555's, and I have 100's of electrolytics on old boards I've scrapped and just in parts bins"... so I didn't get any electrolytics. Now aside from spending an hour and a half hunting high and low for my 555's (which were, of course, in the first place I looked - I just had a man-look as my wife calls it . What I failed to realise was the 2.2uF electrolytic is over the HT side - so needs to at least cover 170V - the spec'd one is 250V.

Anyway - the question. Can you treat electrolytics like little voltage dividers and use, say, 4 10uF 50V in series? Or do they all store the same voltage? My electrical theory is a little rusty so I can't think it through to a sensible conclusion.

Or do I have to stop when I am almost done and wait until tomorrow to buy the right cap (The safe option, of course... but I'm on a roll )



Steve

_________________
<i>There are no personal problems that can't be overcome with the liberal application of high explosives</i>

Connecting electrolytic caps in series is common practice. Variations in manufacturing may affect their internal resistance. To make sure each cap gets 1/4 of the supply voltage, connect equal-value resistors across each one. They can be fairly large. Something like 188k would cause less than 1-mA current through the voltage-divider resistor string. That's four, 47k resistors.

73,

_________________

http://kr1s.kearman.com/
http://qrp.kearman.com/

Thanks Jim... thats going to eat into my 47k supply that I keep handy for parallelling with crystal earpieces... but I think I can live with that

I'll give it a go now and see how it works out (might wear safety glasses just in case... I've been there and done that with electrolytics before)...

Steve

_________________
<i>There are no personal problems that can't be overcome with the liberal application of high explosives</i>

Steve,

47k was only a suggestion, based on having ~1 mA through the divider string. If the supply can handle a little more current, you can use a smaller resistance value if you have more of them. Just remember to do an Ohm's Law calculation. P = V^2 / R If the supply is 170 V and your string total is, say, 200k, the total power dissipated in the resistors is about 0.15 W. Each resistor shares about 1/4 of that (assuming they aren't precision resistors their individual values can vary plus-or-minus 5- or 10-percent). You don't want the resistors warming up very much, as that reduces their resistance, increasing current flow, which further increases heating.... Also, they may not heat equally, unbalancing the voltage divider. I think anything from 33k to 68k per resistor is acceptable.

Those resistors will help discharge the caps on power-off, and also improve voltage regulation, which could be helpful, as the drain on the supply will vary depending on how many segments are lit. But nixies don't consume much current in any event.

73,

_________________

http://kr1s.kearman.com/
http://qrp.kearman.com/

Hi Steve,

what I would recommend is to replace this old IRF740 MOSFET with a modern one. A few years ago I build a nixie clock and the supply circuit was practically identical with the one you linked. There were reports and complains that the MOSFET became really hot and sometimes even explodes - which is not so unlikely since the device can only block 400V. This is not enough for many spikes on the powerlines, at least not here in Europe using a 230V grid. The heating is related to the ohmic and switching losses, and I also realised in my clock that the IRF became rather hot (but do not touch the leadframe if the devices is turned on, you will have the full output voltage at the heatsink of the MOSFET!).

I replaced the IRF740 by a 200mOhm 600V CoolMOS C3 from Infineon (SPP20N60C3), but also the elder SPP20N60S5 will work (I noticed that many distributors rather often offer older types, which ofcourse is sometimes not so bad). Checking the waveforms with an oscilloscope revealed no problems. Just take care not to use the MOSFET having a 'CP' in the name, those devices are designed for really fast applications and are able to interact with your parasitics in the circuit - and those parasitics will be most likely too large in our homemade electronics. There are also similar devices offered by ST or Fairchild and so on.

At least you should add a small heatsink to the IRF740, especially if your clock is once build in an enclosure.


Brösel

Thanks for the tips guys. I actually got the right cap this morning after only being able to find 3 suitable caps when I needed 4 and going to bed in disgust. I've finished it now, but it seems a little unstable. It will flick from 170V to 130V, then drift down around 100mV every second for a while, then stabilize around 125V. There is no obvious pattern - sometimes it works fine, sometimes not.

I've swapped out the FET for a spare, I've swapped out the 555 for a spare and I've swapped the transistor out for a spare (also tried a spare MPSA42) checked the circuit.

The only odd thing is that the resistor seems to vary (from full left to full right) the voltage from 100V up to 135-140 then start down again somewhere near the middle of the sweep.

I have made a couple of changes from the circuit - a BC337 in place of the BC547, a 470uF reservoir cap instead of 330uF and a BYV26C - 600V 1A diode in place of the UF4004.
[edit] I actually think its topping out at 140ish volts then its just a little random what it provides after that. And I noticed the nixie pulls the HT output down by about 3V - and since it should only be pulling around 1.5mA that seems a little excessive... or am I worrying about nothing on that score? [/edit]

The other thing I have noticed is the voltage drifts all over the place (+/- 10V or so). I guess this isn't too critical for a nixie, but its worrying me that the supply isn't the best choice. I may go for a DC-DC converter-based circuit instead unless someone can spot what might be wrong... unfortunately I don't have a scope to check waveforms around the timer chip...

Any thoughts? Or should I just get back to crystal sets? I got a few BAT48 diodes to try in place of my OA91's while I was getting the parts for this

Steve

_________________
<i>There are no personal problems that can't be overcome with the liberal application of high explosives</i>

I have built this same circuit with reasonably good results, though I wanted only 60 V DC out to power a one-tube radio.

1. What is the source of your input DC? When you raise the output voltage, you have to pull more current from your low voltage source. If the current can't be provided, the input voltage drops and then the circuit goes unstable.

2. An IRF740 is actually overkill in this circuit. No need to worry about mains voltage as this circuit doesn't operate with a mains input. The 740 has a large input capacitance and the 555 has to supply extra current to charge and discharge on every cycle. An IRF720 type should be more than enough.

3. The snubber (R/C across the MOSFET) needs to be wired close to the FET and the capacitor should be mica or ceramic. Have you put a scope across the MOSFET to see if you have any parasitic oscillation or ringing (could be due to circuit lay-out).

4. How much current are you drawing on the hi voltage side during your tests?

5. I had much better stability and regulation with the MAX1771 boost converter IC.

http://www.desmith.net/NMdS/Electronics/NixiePSU.html

Rich

_________________
Homebrew Radio ex-Silicon Valley

Is this with a load on the output, or no load?

A problem with that 555 power supply circuit is that it's not really regulated in the normal sense of the word. When the output voltage goes above the setpoint, the supply shuts down completely and then restarts when the voltage drops below a certain point, which isn't going to give very good regulation.

A properly regulated switchmode supply uses a continuously variable pulse width to the MOSFET to give constant output that immediately adjusts for variation in load.

The feedback circuit in the 555 circuit doesn't have any time dependent components that could be tuned for better response. However, it may work better when there is a reasonable load put on the output.

Edit:
It may also help to try a different NPN transistor. Maybe yours is borderline.

Actually discontinuous mode regulators are used pretty frequently.

http://www.infineon.com/cms/en/product/ ... 6a73f6050c

More info is needed to completely diagnose the problem with the 555 circuit. It can be made to work pretty well. Certainly well enough to power a Nixie clock. I doubt you will see a brightness difference if the voltage varies by 1 or 2 volts.

Rich

_________________
Homebrew Radio ex-Silicon Valley

I guess so, but I don't have to like it.

The question here, is whether the 555 version is sufficiently responsive. For example, the 555 control input may have a bit of hysteresis, which could cause the output voltage to cycle up and down.

Regardless, that 555 circuit has been around for a while, and is quite popular. So, I assume that it must work well enough for the application.

Bob,

I don't disagree. The 555 was designed well before switch-mode regulators were popular and isn't optimized for that job. It can be made to work "well enough" but I like the more modern application-specific circuits. Depending on the load and your circuit lay-out, you can get all sorts of oscillation and instability in a boost circuit.

The biggest thing to remember is that if you want high voltage from a low voltage source, the current from the source has to be pretty high. I was originally thinking of running a boost circuit from 4 x AA alkaline to give me 60 V at maybe 20 mA. That's only 1.2 watts, but it means over 200mA average out of the batteries and peaks can be up to an amp.

Rich

_________________
Homebrew Radio ex-Silicon Valley

Hi guys, sorry but I lost track of this thread. I moved the snubber circuit to get it close to the FET as suggested, but this made no difference. I don't have a scope (although I am bidding on one at the moment, so maybe that will change in a few days).

Its powered from a new alkaline 9V battery, so the current supply shouldn't be an issue. I tried another battery, same deal.

I've also tried 3 different 555's now.

Its been on hold for a while due to workload, and with a day to myself the other day I passed it over in favour of playing with my crystal set (mainly because it works too well, as opposed to the power supply, which doesn't work well enough )

Once I get a scope I will look at it again... cheers for the advice - I'll keep you posted.

Steve

_________________
<i>There are no personal problems that can't be overcome with the liberal application of high explosives</i>

I think it is an issue. If you are trying to get 170v at 20 mA output, that means you need at least 9V at 370 mA input! That's if the circuit was 100% efficient.

9V batteries are good for projects under 100 mA. At 300 mA, the internal resistance of the battery is going to lower the voltage and cause all kinds of instability.

Have you tried this circuit with a power supply capable of at least 500 mA?

If you had a scope, you might see that the peak current in the inductor rises to maybe one ampere. A 9V battery isn't going to cut it.

Try 6 AA cells in series; better yet, 6 x D cells.

Rich

_________________
Homebrew Radio ex-Silicon Valley

Check out a typical 9V battery data sheet:

http://www1.duracell.com/procell/pdf/PC1604_US_PC.pdf

1. The graphs show a worst-case load of 100 ohms; your load is more like 10 ohms or less.

2. Even at 100 ohms, the battery voltage falls almost immediately to about 8 volts; that means your converter will try to make longer pulses to keep the output voltage regulated, resulting in even more power sucked from the battery.

3. The internal resistance is 1.7 ohms. at 300 mA current draw, the battery voltage will be about 8.5 V with a fresh battery. If your circuit is like mine, the peak inductor current is about an amp, so the battery voltage will sag during the pulses to about 7.3 V, resulting in very poor regulation of the output voltage.

Let us know how it works on a different power source.

Rich

_________________
Homebrew Radio ex-Silicon Valley

Here is a nifty calculator for simple boost converters.

http://www.ladyada.net/library/diyboostcalc.html

I ran one example of 9V in and 180V out at 20 mA.

The peak inductor current is about 0.8 amps. That's got to be sourced from your battery. A 9V alkaline won't do it.

Rich

_________________
Homebrew Radio ex-Silicon Valley