I am looking for differential delay cell with wide tuning range. The delay range is from 30/40~60/70ps. Target is 50ps!

I tried to use simple differential cell with pmos active load + replica bias (Maneatis type without diode-connected transistor). I found the range is too small to cater process variation..

thinking for programmable load or mux...
any advice..?
any good high-speed delay cells?

yes. the amplitude is getting smaller in some range.

I notice there are two types of famous differential delay cell.

1) Maneatis type -->adjusting Vctrl (amplitude), changing I
2) Maneatis type -->adjusting Vbias (current), changing V  

which one is better?
The only difference i found is the tuning range. 2) tends to be smaller range but frequently used in DLL

Yes. Maneatis delay cell is the most wide range i have ever found.
However the wide range is mostly cover the low freq range.

50ps delay and the input freq is 10Ghz. The amplitude is getting small at high freq.

Code:

what do you mean by '2) Maneatis type -->adjusting Vbias (current), changing V  ' 

Some designs do not use opamp, directly adjust vbias (Vbn) of the current source & replica bias.  You could refer to those DLL papers.

A Wide-Range Delay-Locked Loop With a Fixed Latency of One Clock Cycle

Basically it is similar to maneatis type but it is without opamp. It is suitable for DLL application since DLL reset from max freq to target freq.

I need to cover process variations and i am using old process 0.18um. Pushing to the limit of 10GHz.

I manage to get it but the power consumption is relatively high.