How to get different AC voltages - AC voltage regulator?

[ds18s20]

Hi everyone

So this is a retarded question yet I cant get a straight answer from Google

I have never worked with AC stuff other than RF but I would not count that as AC since many things are different in high frequencies. What I am talking about is good old 50Hz 500mA source - I need to know if resistor network will divide it just as it would in the DC world and also when one runs the numbers - is the norm to use rms for calculations and all intensive purposes?

I also need to know if there is such thing as adjustable AC voltage regulator - I just need to have few different AC voltages for this weird display and I cant figure out how to produce them - all I have is DC gear?

 

[lambtron]

What I am talking about is good old 50Hz 500mA source - I need to know if resistor network will divide it just as it would in the DC world and also when one runs the numbers - is the norm to use rms for calculations and all intensive purposes?

Yes, a resistive divider will work just as in the DC world at low frequencies.

I also need to know if there is such thing as adjustable AC voltage regulator - I just need to have few different AC voltages for this weird display and I cant figure out how to produce them - all I have is DC gear?

Google for "variac" or "variable autotransformer."

 

[ds18s20]

Google for "variac" or "variable autotransformer."

Thanks, I found ton of them. However those are mechanical devices, isn't there something semiconductor, like the 2-3A DC voltage regulators that we are all used to in the DC world?

That would be ideal as the variacs are huge for pcb integration?

Thanks
B

 

[lambtron]

What are the display's requirements? That is, what AC voltages does it need, and at what currents and frequency? Are they all fixed voltages?

 

[ds18s20]

What are the display's requirements? That is, what AC voltages does it need, and at what currents and frequency? Are they all fixed voltages?

Hi, yes they are fixed. Fillament is 3.5V AC at 300mA, grid and Anode are at 20V DC and 50mA, and theire is a requirement for DC bios at the center tap of the AC transformer like such:

**broken link removed** **broken link removed**

Here is what the general VFD driving guide says:

**broken link removed**

Thanks for your help
Boyan

 

[lambtron]

It's pretty clear that Noritake has in mind a center-tapped transformer for driving the filament. There are probably lots of other ways to power the filament, but two come two mind right away:

1. If you have a microcontroller nearby, have it generate a 0-5VDC sine wave at 50-60 Hz, then buffer and level-shift that with two power op-amps such as TI's OPA569. Configure both op-amps to output the sine wave centered about 0V, but their outputs will be 180° out of phase with each other. Each op-amp drives one side of the filament. There is no need for "center tapped" transformer equivalent because both op-amp outputs are ground referenced. Digitally generating the sine wave is simple with a lookup table and an inexpensive dac or, even cheaper, pwm on a digital output with low-pass filter.

2. Noritake says you can use direct digital drive for the filament, but they are short on details and insist that you contact them to discuss your drive design with them before implementing it. It's possible that this would be a simpler and less expensive solution, but that depends a lot on how communicative they are.

Hope this helps!

 

[ds18s20]

Hi,

Thanks, you mean hook up the op-amps like this:

**broken link removed**

 

[BlakFlame]

umm i dont know what the circuits in the pics does..
but i'll answer the question about how to get diff AC voltage..and the regulator Q

if u use Op-amps with the proper feedback you'll get all u want!..

1. you'll get all the ac Voltage signals u want.

2. Ideally the op-amp has a low output resistance. which means the AC voltage coming out of it IS regulated.

now wether to use Power Op-amp or normal ones, that depends on the voltage and current ratings you need ..

hope thats clears thing up ...:|

 

[Mr Cappuccino]

A transformer tapped at different points would act as a perfect AC voltage regulator.

If resistor combination is to be used then loading effect might be pronounced .... I would prefer using a transformer tapped at different points on the secondary ... an autotransformer can be used as well

 

[lambtron]

... you mean hook up the op-amps like this:

No, that won't work because the opamps are running open loop.

See the attached diagram. This shows how to use two pwm signals, 180° out of phase with each other, to generate a differential 3.5VAC after passing through low-pass filters. The filters are followed by power opamps, which in turn directly drive the filament.

I specified a nominal 1KHz pwm frequency, but this is somewhat arbitrary. It should be much higher than the 50Hz output frequency you are deriving from the pwm, and 1KHz meets that criteria.

The low pass filters probably don't need to be "perfect" as the filament has some inductance that will help to smooth out digital edge effects that are not blocked by the filters.

Note that the AC waveforms have a +1.75V bias because their negative peaks bottom out at 0V. Because of this, you may need to increase the anode voltage by an equal amout, from 20V to 21.75V to keep the display happy.

I hope this helps.

 

[ds18s20]

phase with each other, to generate a differential 3.5VAC after passing through low-pass filters. The filters are followed by power opamps, which in turn directly drive the filament.

The low pass filters probably don't need to be "perfect" as the filament has some inductance that will help to smooth out digital edge effects that are not blocked by the filters.

Note that the AC waveforms have a +1.75V bias because their negative peaks bottom out at 0V. Because of this, you may need to increase the anode voltage by an equal amout, from 20V to 21.75V to keep the display happy.

Hi,

Thanks, that schematic makes a good point. Just one question - what's the math behind it? How did we get a 3.5V amplitude out of a 5V square wave signal which comes out of the micro just by passing via the filter? Then how did we get 50 msec period out of a 1kHz square wave, again just by passing via the filter. Lastly how do you figure 1.75V bias?

Ahh and the output AC - why isn't that 7V p-p, if we have each composite amplitude go to +3.5V? The output should see 7V p-p wouldn't it?

Thanks
~B

 

[lambtron]

How did we get a 3.5V amplitude out of a 5V square wave signal which comes out of the micro just by passing via the filter?

The low pass filter "averages" the pwm output. If you have a 50% duty cycle on the digital output, the filter output will give you 2.5V. And at 70% duty cycle the filter will output 3.5V (3.5/5.0*100 = 70%). So, the pwm output will "sweep" back and forth between 0% (0V) and 70% (3.5V).

Then how did we get 50 msec period out of a 1kHz square wave, again just by passing via the filter.

You program the pwm output to produce pulses whose widths sweep between 0 and 70% duty cycle at a rate of 50Hz.

Lastly how do you figure 1.75V bias?

Both of the pwm filters output a sinusoid that ranges from 0 to 3.5V in amplitude. The voltage about which the sine wave is symmetrical is the average voltage, which is (0+3.5)/2 = 1.75V.

Ahh and the output AC - why isn't that 7V p-p, if we have each composite amplitude go to +3.5V? The output should see 7V p-p wouldn't it?

Well think about it: when one output is 0V the other is 3.5, and vice versa. So the differential amplitude never exceeds 3.5V. And the differential amplitude is what is seen by the filament.

 

[ds18s20]

I got it finally,

So my 5V square wave will be modulated as such so when fed via the low-pass filter things will be smoothed out exactly as we want. Also thanks for the clarification on the math - I now understand how you arrived at the numbers. Ahh and I can't do 0% duty cycle PWM - best I can do is 1% - just software limitation but I am sure 1% will not screw up much as it is soooo quick at 1kHz - a period of 1 mSec/100 is 10 micro seconds so 1% duty cycle is 1 micro second raise.

**broken link removed**

 

[lambtron]

Yes, that's correct. The output pulses occur at a fixed frequency of 1KHz, but the pulse widths are varied to give you the desired duty cycle.

You can use various different methods to compute the pulse widths so as to generate a sinusoid. Some of these are:

* If firmware memory is plentiful and you don't want to think about it too hard, use a lookup table. Just keep cycling through the table, one table pass per sinusoid cycle.

* If firmware memory is scarce or you want to be clever, there are various compact algorithms available to compute the pulse widths on the fly.

Glad I could help you!