Nixie SMPS efficiency tested

[gelbanana]

Today I've decided to test the efficiency of the different design of HV SMPS that I've built a while back. I used a 240v neon bulb which I salvaged from a panel mount light indicator. The bulb draws about 1.5mA and I've used 2 pieces. MAX1771 have the best efficiency but comes with a pricey price tag.

[matseng]

Nice units. Can you repeat the test and trim the voltages on all units and both low and high loads to have the same voltage? Like 180 or 200 volts.

The output drops quite a lot when loaded - a load change of 0.9 mA causing the MAX1771 to drop 40 volts seems iffy to me. Either you have adjusted the voltage manually between the tests or the smsp's are running very close to their margins. Do you use a proper coil? All three designs do have a feedback loop which should keep the voltage within a just a few volts between 0-load and a few milliamps.

[gelbanana]

Well I've built them based on the reference design and set all the pots to max. Could be that each have their own min-max range. Looks like i'll have to bring out the voltage-divider-fu to see what's their ranges are. I recalled a while back i used a 555 design to drive 6 IN-8 tubes. Voltages drops from 202v to 136v@4.68mA.

I'm guessing you are right on the coils. I ordered these on E14 which which they labeled 100uH, 670mA but I received '105' coils, 1H I supposed?
http://sg.element14.com/murata-power-so ... t=107-7029

Will test them again when I got the right coils

[erdabyz]

You can get an efficiency estimation of a SMPS design with your method, but if you really really want to measure true efficiency you have to do something a little more complex. Basically SMPS efficiency is measured by setting load conditions, waiting for a steady state (ie. when the behaviour of the voltages and currents involved in the circuit are mostly periodic) and then integrating power in both input and output over several cycles of that periodic behaviour. If your multimeters are true RMS (which I suppose as they are flukes) and behave well at the frequencies involved in the SMPS design then the estimations should be fairly accurate.

Anyways, the efficiencies you report are quite low. I'm not sure about how MAX1771 works but MC34063 and NE555 designs should be straightforward PWM and with such a control method efficiency tends to be low at light loads. Probably MAX1771 incorporates some sort of "burst mode" behaviour, as does LT1073 for example, which dramatically improves light load efficiency.
You could also have some problems with the MOSFET behaviour. MC34063 (and likely NE555) designs use bipolar transistors and (at least MC34063 ones) drive the gate of the MOSFET from the emmiter of the output bipolar transistor of the controller, and rely on a "bleeding" resistor to drive the gate low. That's OK for low frequency switching, but at low frequencies the inductors need to be large. For example, my 6 tube nixie clock has a MC34063 based PSU design and it uses a fairly large 330uH inductor which gets slightly warm over time meaning that appreciable losses are occuring in there. Haven't measured efficiency. Maybe using a larger inductor and a mosfet with less gate capacitance helps.

[gelbanana]

Good point erdabyz. I know the 555 outputs at 32kHz, fixed 50% duty cycle, the MC34063 40kHz not sure if it does pwm. Switching speed seems ok with the the bleeder resistor used.
Anyways I've measured the inductors and they are 1mH instead of the suggested 100uH. Could this be the cause of the poor efficiency?
Any thoughts on using lower Vf diodes? Since they are the main loss in a dc-dc converter. The UF4004 is commonly used and it has a max Vf of 1v. I've searched in E14 and the lowest Vf max fast recovery diode is 0.82v.

[erdabyz]

Good point erdabyz. I know the 555 outputs at 32kHz, fixed 50% duty cycle, the MC34063 40kHz not sure if it does pwm. Switching speed seems ok with the the bleeder resistor used.
Anyways I've measured the inductors and they are 1mH instead of the suggested 100uH. Could this be the cause of the poor efficiency?
Any thoughts on using lower Vf diodes? Since they are the main loss in a dc-dc converter. The UF4004 is commonly used and it has a max Vf of 1v. I've searched in E14 and the lowest Vf max fast recovery diode is 0.82v.

My nixie clock uses a BYV27-400 as the switching diode. It's an ultra fast recovery power schottky diode. The -400 version is quite hard to find, they usually have the -200 and the -600. I found mines at futurlec. 1mH looks like too much inductance, but the real problem comes with the DC resistance. Being the size they are, those inductors will likely have a lot of turns with small diameter wire, causing them to be fairly resistive. That's not good for a DC-DC converter. Maybe you can look at coilcraft. They have proper power inductors with very low DC resistance and they send free samples of them.

[gelbanana]

I've requested some samples from coilcraft. Also ordered some murata ones from E14. Will test them again and see how it goes

[gelbanana]

Small update. I've replaced the inductors with the correct values. The voltages on the 555 seems to be stable maintaining at 200v on 2.5mA load.
The voltages on the MAX1771 n MC34063 seems to over shoot past 300v, which forced me to stop further testing. The output cap got really hot which might almost ruptured. I'll have to check the voltage dividers. Will probably do more test this weekend.

[markeprice]

I am using a nixie type power supply that uses a 555, IRF740 and BC547c. I am having a terrible time getting anything other than the 12-15VDC at the input at the output of the high voltage (B+). Basically if I have 12VDC going in, I have 12VDC at the output. I have swapped out all the compenents listed earlier plus the 4.7uf cap. No change. Has anyone got any troubleshooting tips for me? I have gone over a few different schematics that use the same or similar components and cross checked my pcb implementation (on a superfly amp board by Rick Holt, frequency central) and cannot see what might be the problem. I am guessing since there is no difference between the input and output that there is no "charge pumping" going on. I don't own a ocilliscope, just a multimeter. Can you provide a recent tested and verified schematic of this type of supply. Thank you. Getting frustrated! Mark

[matseng]

I built a copy of this http://www.ledsales.com.au/kits/nixie_supply.pdf for the 7400 competition viewtopic.php?f=62&t=4735. Works just fine even with slightly different components.

Nixie smps for 7400 competition