The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> How to design current source for RF circuits??
https://designers-guide.org/forum/YaBB.pl?num=1213240382

Message started by zhangjerome on Jun 11^th, 2008, 8:13pm

Title: How to design current source for RF circuits??
Post by zhangjerome on Jun 11^th, 2008, 8:13pm

Generally we use long channel devices to design current sources in low frequency circuits. Long channel device leads
to large parasitics, which can be significant in RF application. Is there any rule of thumb for current source design??

Thank you.

Title: Re: How to design current source for RF circuits??
Post by vivkr on Jun 12^th, 2008, 3:24am

which parasitics are we talking about?

Gate-drain cap (CGD) is mostly overlap cap and thus independent of L.
Gate-source cap (CGS) should not matter as it will only load the bias reference levels from a diode connected transistor.

Longer L will actually give you a good current source and better PSRR, well not at RF though :)

Regards
Vivek

Title: Re: How to design current source for RF circuits??
Post by Hyvonen on Jun 12^th, 2008, 8:40am

Zhangjerome: Do you mean that long-channel device leads to high parasitics because you have to increase also the width to get the same W/L (or overdrive for a given current)?

You will have to just take the hit in either Rout or parasitics... or, if you have headroom, you could consider using a cascode. And Vivek makes a good point - at what frequency do you want your PSRR to be high (and how high)?

Title: Re: How to design current source for RF circuits??
Post by zhangjerome on Jun 14^th, 2008, 1:27am

Yeah, that's what I mean. Both W and L have to be increased if long channel devices are used to implement the current source with constant current.

It's not only the PSRR. Small output impedance and large parasitic cap both lead to RF signal leakge if the current source is used as load.

Title: Re: How to design current source for RF circuits??
Post by Hyvonen on Jun 14^th, 2008, 1:16pm

I assume you don't need to match the output impedance to 50ohms etc.
If your amplifier is narrow-band, maybe an LC tank could do the job. If it's broadband, then maybe you could use an actual physical resistor as a load...? Or are you really shooting for a maximum possible gain (like in an opamp), while maximizing gain? I think I would look into a cascode in that case.

Title: Re: How to design current source for RF circuits??
Post by loose-electron on Jun 16^th, 2008, 12:34pm

One other thing - Current sources are noisy! A lot of engineers design RF systems using current sources and active load structures --- until --- they look at the noise performance! Then, things get redesigned into resistors, or LC methods.

Also, don't obsesses on keeping the currents perfect and put cascodes everywhere. In most cases the current can vary some and you dont need the cascode. More headroom, fewer components and usually quieter