Wideband AC Coupling for DDS Outputs

[Jdo300]

Hello All,

I've been playing worth some DDS output filter and amplifier circuits in the simulator and have run into an interesting little problem. I want to make an output that can range from 0.2Hz to 20Mhz and so far I have built a very nice reconstruction filter that can give me a very clean sine wave through that whole range. The problem is that I want to also have offset and amplitude adjustments for the output. I am going with a couple of high speed op-amps with digital pots for the adjustments but things still get complicated because at the high frequency end (10-20Mhz) where the filter begins to attenuate the signal a bit, the offset changes. Also, I have a simple digital pot on the end of the filter with a center tap that I use as the amplitude adjustment. When ever that changes, that effectively changes the DC offset of the signal as well. I was thinking that in the absolute worse case, I could include all these factors in the programming of the microcontroller so that it can compensate for it. But I'm wondering if there is a better way to deal with this problem?

I know it is common to just use a capacitor in series with the output going to the op-amp but i have tried this and it doesn't seem to work so well at very high frequencies or very low frequencies (I tried a 1uF cap with a 47Ohm resistor in shunt to ground). My design does use a bipiolar supply so I'm wondering if I could somehow use this to my advantage. The DDS raw output sine wave is 0 to 600mV though that can vary a bit after it goes through the filter circuit. What do you all think?

Thanks,
Jason O

 

[waht]

I know it is common to just use a capacitor in series with the output going to the op-amp but i have tried this and it doesn't seem to work so well at very high frequencies or very low frequencies (I tried a 1uF cap with a 47Ohm resistor in shunt to ground).

it's possible to connect capacitors of various values low and high in parallel. That will give you the effect of a wide band coupling. As one capacitor is increasing in reactance, the parallel one can have lower.

 

[jambaugh]

I'm no circuit Guru, I only know theory so I'm only good for ideas not experience...

Having said that, while reading my first thought was the series cap which you then mentioned. Clearly for low frequency operation you'll need a high capacitance and the problem at high frequencies is that the cap is not ideal and has some inductance (worse the higher the capacitance).

Some off the cuff suggestions:

Can you find a virtual capacitance circuit using op amps?

What about using an array of high performance caps (suitable for the high freq. domain) connected in parallel (reduces inductance while increasing capacitance) with carefully equal lead length to prevent mixing phases? May be cost prohibitive but it seems you could up the capacitance that way without degrading HiFreq performance.

Hmmm... OK I have a signal in a line relative to ground and I want to output it with independent amplitude and offset adjustments. For amplitude control I would use a pot between signal and ground with tap to the positive input of an op amp.

In addition I'd connect the negative input of the same op-amp to the center tap of a separate pot of high resistance connected across my supply and ground. (High enough resistance so the pot doesn't become an electric heater!) That plus something to keep the gain at unity would (in theory) give you independent control of amplitude and offset.

Did that make any sense? Let me try a diagram:

That should work in principle with the amplifier acting as a unity (or higher if desired) gain difference amplifier. If not maybe it will spark some better ideas.

 

[jambaugh]

it's possible to connect capacitors of various values low and high in parallel. That will give you the effect of a wide band coupling. As one capacitor is increasing in reactance, the parallel one can have lower.

Yah! that's much better than my array. (Was drawing while you posted).

 

[Jdo300]

Hi Everyone, Thank you for your help with this problem. I think I have found a really good solution in my case. The DDS IC that I am using, the AD9834, has a current output DAC with differential outputs. Originally I was going to use it single-ended with my filter circuit and op-amp but after doing some further thinking, I realized that if I simply add a filter for each output and run them both differentially into the op-amp, I can totally eliminate the offset problem all together . I tested out the final solution and it seems to be working flawlessly.

Thank you all for your help and suggestions though. I'll definitely keep your ideas in mind for my next project.

Thanks!
Jason O