The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> question about polyphase filter
https://designers-guide.org/forum/YaBB.pl?num=1184076703

Message started by James Bond on Jul 10^th, 2007, 7:11am

Title: question about polyphase filter
Post by James Bond on Jul 10^th, 2007, 7:11am

Hi, all

I built a polyphase filter for 1.9GHz. The resistance at the first stage is 500ohm, at second stage is 1000ohm, capacitance is caculated according to C=1/2*pi*f*R. The sinsoid signal is generated from vco.

Title: Re: question about polyphase filter
Post by James Bond on Jul 10^th, 2007, 7:28am

The polyphase schematic is like this:[img][/img]

I don't know why the Q branch is heavily distorted. Can anybody tell me the reason ? Thank you very much.

James

Title: Re: question about polyphase filter
Post by James Bond on Jul 10^th, 2007, 7:31am

sorry for the mess. curve2

Title: Re: question about polyphase filter
Post by ACWWong on Jul 10^th, 2007, 8:23am

I think you must have loaded the vco incorrectly... perhaps you should look at the input signal to make sure it looks ok.

Title: Re: question about polyphase filter
Post by aaron_do on Jul 10^th, 2007, 8:35am

just thought i'd mention that i think if you connect the input as ++-- instead of +gnd-gnd you can get a cleaner signal and all the output DC levels will be the same, however there is more loss.

Title: Re: question about polyphase filter
Post by James Bond on Jul 10^th, 2007, 1:34pm

Hi, ACWWong and aaron_do, thanks for your replies.

to ACWWong: I checked the vco output , it's not a perfect sine, but it's clean, even just before the polyphase filter, the input is still clean.

to aaron_do: could you please explain "if you connect the input as ++ -- instead of +gnd - gnd you can get a cleaner signal and the output dc level will be the same" in details ? dc level are not the same is also another question, thanks for reminding me.

Thank you very much.

Title: Re: question about polyphase filter
Post by James Bond on Jul 10^th, 2007, 1:36pm

To aaron_do: do you mean I should change upper gnd to Vi+ and lower gnd to Vi- ? So actually, the upper two inputs are all connected to Vi+, the lower two inputs are all connected to Vi- ?

Title: Re: question about polyphase filter
Post by James Bond on Jul 10^th, 2007, 1:51pm

hi, I just checked aaron_do's idea, now the output of polyphase is like this :[img][/img]

I think maybe I need a buffer amplifier before the polyphase, even the input looks ok, it might still load the vco output. But another thing is why it only load one branch ? the Vi_Q branch is distorted , but vi_I branch looks ok, why?

Can anybody give me some suggestion about how to design a buffer amplifier?

Best Regards
james

Title: Re: question about polyphase filter
Post by James Bond on Jul 10^th, 2007, 1:53pm

the schematic is like this:

Title: Re: question about polyphase filter
Post by ACWWong on Jul 10^th, 2007, 2:56pm

have you tried just driving your initial polyphase circuit with an ideal sine voltage source and no loading?

If the outputs are not nice quadrature sine waves (after a short settle time) then there must be something wrong with the R and C components you are using. At 1.9GHz your parasitic capacitance of the large resistors could come into play, as would your capacitors if they are MOS based. Perhaps try using analogLib res and cap first to make sure.

Also when designing two stage polyphase on chip, its often the case that the two stage's frequency are offset slightly to improve IQ amplitude imbalance in the presense of process variation.

Title: Re: question about polyphase filter
Post by James Bond on Jul 10^th, 2007, 11:25pm

Hi, ACWWong,

Yes, I have tried using a ideal sine wave to drive the polyphase and it works, the output is good sine wave.

Is it because of absent of the buffer amplifier ?

Title: Re: question about polyphase filter
Post by aaron_do on Jul 11^th, 2007, 12:36am

Hi,

sorry i couldn't get back to you before but you connected it like i suggested anyway...

I noticed that you said Q is distorted but I is ok....actually both of them look distorted to me. Following that fact, are you positive the input signal is not distorted?

BTW you can do an AC analysis on the poly-phase filter to see if the response is ok. Input two Vsin of opposite phase and look at the 4 outputs. That way you can also see the usable bandwidth of the PPF.

Aaron

P.S if the input looks ok then you don't need a buffer...

Title: Re: question about polyphase filter
Post by James Bond on Jul 11^th, 2007, 2:16am

Hi, aaron_do,

please look at the curve.jpg, the green one is the Vi+_I, the blue one is Vi+Q , both of them are from Vi+.

The Vi+_I looks quite close to the vco output. you can say my vco output(polyphase input) looks like this. however, Vi+_Q looks quite bad. That's what I mean. Is this vco output ok to use?

curve5 is the output with perfect sinesoid source:
curve6 is ac respond:

Title: Re: question about polyphase filter
Post by James Bond on Jul 11^th, 2007, 2:17am

curve6

Title: Re: question about polyphase filter
Post by James Bond on Jul 11^th, 2007, 2:22am

it seems that the problem is from my vco , I don't believe this, anyway, I will post my vco output curve(it's also the input of polyphase filter, before I have a two big caps to remove the dc voltage):

the output of vco start to oscillate stably at 7ns.

Title: Re: question about polyphase filter
Post by aaron_do on Jul 11^th, 2007, 3:10am

hmm...i can't really figure it out. Your VCO output seems a little distorted by maybe that's just my eyes. Perhaps you could use the calculator to do FFT on the VCO output and the polyphase output to see the spectrum. Also for the PPF AC analysis, i meant you should observe the phase of all 4 outputs and make sure they have a 90 degree difference at the operating frequency.

I am not really an expert on VCO design...hope you can figure it out.

Aaron

Title: Re: question about polyphase filter
Post by James Bond on Jul 11^th, 2007, 4:43am

Hi, ACWWang and aaron_do,

I think I solve the problem. It's because of vco output is distorted too much. Now I retune the varactor and transistor to reduce the distortion, the ouput at the polyphase filter looks much better than before. curve8: